1. Field of the Invention
The present invention relates to bipolar junction elements like a junction diode or a junction transistor and a method for manufacturing thereof. Especially, the present invention relates to a bipolar transistor and a method for manufacturing the bipolar transistor in which a silicon or a silicon-germanium(SiGe) film is used for a base.
The field of semiconductor with highly advanced technology can be divided into the field of memory, the field of system integrated circuit(IC) which is represented by application specific integrated circuit (ASIC), the field of radio frequency integrated circuit (RFIC) which is essential in wireless communication, and the field of high speed digital and analog integrated circuit for data process. Among these fields of semiconductor, the present invention is especially related to the semiconductor element having high maximum oscillation frequency(fmax) and high cut-off frequency(fT) which are related to the field of RFICs and the high speed digital and analog ICs. As we live in the era where wireless communication is widely used and where its need is increasing explosively, the need of high frequency elements is rising as the quality of communication improves and the markets of higher frequency band providing various communication services and large number of subscribers increase. Also, as the need of super speed information communication network increases in the contemporary society, researches and developments of a high speed transistor are carried on actively. At present, 10 Gbps integrated circuit using high speed transistor about 60 GHz is developed in the usage of optical communications and is at the stage of being commonly used. The silicon homo bipolar junction transistor which includes the silicon base layer formed by ion-implanting a dopant has maximum fT of 30 GHz at most. The heterojunction bipolar transistor(HBT) having an epitaxially grown SiGe base layer exhibits maximum fT in the range of 50xcx9c150 GHz and fmax of 50xcx9c160 GHz.
The SiGe has a narrower energy bandgap(Eg) than that of the silicon. The energy bandgap difference(xcex94Eg) between the silicon emitter and the SiGe base enhances the current gain exponentially, and the fT and the fmax also increase compared to the homojunction bipolar transistor. Therefore, the impurity doping concentration into the SiGe base can be increased in the margin of exp (xcex94Eg) without degrading the current gain. The base resistance is lowered and consequently the noise figure is lowered. Furthermore, it is possible for power consumption to be lowered because the bias current achieving the same fT decreases. In comparison with the base of the transistor formed by ion-implanting in the conventional method, the base formed by the epitaxial growth method can be much thinner to the degree of 200 xc3x85 approximately and consequently the cut-off frequency increases. Since the Ge composition in the SiGe base from the emitter side to the collector side ramps up linearly, the electrons transiting to the base accelerate. Therefore, the fT and fmax can be further increased by grading the Ge content. The SiGe HBT is fully process-compatible with silicon devices. Furthermore, it allows to achieve higher fT and fmax than 100 GHz using 0.8xcx9c1 xcexcm of photolithography. This means that, contrast to memory and ASIC getting highly scale-down to 0.18xcx9c0.25 xcexcm, SiGe HBTs can be fabricated by recycling the out-of-date production facilities at the 0.8xcx9c1 xcexcm level. Therefore, it has good economic value with high output.
2. Description of the Prior Art
There are several registered patents showing related arts of SiGe HBT from IBM in the United States, NEC, Hitachi, and SONY in Japan, TEMIC in Germany, and Electronics and Telecommunications Research Institute(ETRI) in Korea. The structural characteristics and drawbacks of the related arts will be given in the followings.
First, the prior art of NEC in Japan is a kind of the super self-aligned NPN HBT. In this particular transistor, the base layer including SiGe is selectively grown in the device active region and each of the collector-base and the emitter-base junctions is self-aligned. The method for manufacturing this super self-aligned will be described herein below.
In FIG. 1a, a n+ type buried collector 11 is formed by ion-implanting n+ type impurity (dopant) into a pxe2x88x92 type silicon substrate 1. A collector layer 10 is deposited on the resulting structure. A collector sinker 13 which connects the buried collector 11 and a collector semiconductor electrode to be formed afterward is formed by implanting n+ type impurity ions into the region as shown in the figure. A trench is formed by etching the collector layer 10 and the substrate 1 in order to isolate the neighboring transistors electrically. The isolation trench 71 is filled with an insulation material like boron phosphorous silica glass (BPSG). Then, the surface of the isolation trench 71 is planarized by a chemical-mechanical polishing (CMP) of the BSPG so that the surface of the isolation trench 71 becomes a same height with the surface of the collector layer 10. Form a collector insulation layer 17 with silicon oxide layer, a base semiconductor electrode 23 with p+ type polysilicon film, and an emitter insulation layer 37 with silicon nitride film by sequentially depositing on the substrate 1 where the collector layer 10 and the isolation trench 71 are formed. The collector insulation layer 17 in the region which is planned to be the emitter is exposed by etching the emitter insulation layer 37 and the base semiconductor electrode 23. Then, by depositing an insulation layer and etching it anisotropically, a first side-wall insulation layer 73 is formed at the inner side-wall of the base semiconductor electrode 23 and the emitter insulation layer 37. Wet etch the exposed collector insulation layer 17 in order to expose the collector layer 10 beneath the collector insulation layer 17. Even after the collector layer 10 is exposed, continue the wet etching to form an undercut 27a to the pre-determined depth beneath the base semiconductor electrode 23. The n type impurity ions are added selectively to the intrinsic collector region 15 by ion-implanting to the resulting structure in order to increase the cut-off frequency.
In FIG. 1b, a base layer 20 composed of an undoped SiGe, a p+ SiGe layer, and an undoped Si layer which is supposed to be the emitter 35 later on is grown selectively on the exposed collector layer 10 and beneath the exposed base semiconductor electrode 23 in the undercut 27a. Here, a base connecting part 27b which is selectively deposited beneath the base semiconductor electrode 23 is a poly-crystalline while the base layer 20 on the collector layer 10 is single crystalline. A silicon film is further selectively grown thereon in order to make sure the connection between the base semiconductor electrode 23 and the base layer 20. At this step, growth rate of the single crystalline silicon layer on the base layer 20 is controlled to be much slower than that of the poly-crystalline base connecting part 27b, so that the thickness variation of the undoped Si layer at the top of the base layer 20 is minimized. The second side-wall insulation layer 75 which covers the first side-wall insulation layer 73 and which contacts with a part of the base layer 20 is formed by depositing the insulation material like a silicon nitride film and by etching it anisotropically. Then, expose the collector sinker 13 by partially opening the collector insulation layer 17. An n type polysilicon layer is deposited on the resulting structure. Then it is patterned to form an emitter semiconductor electrode 33 on the base layer 20 and a collector semiconductor electrode 13a on the collector sinker 13. Diffuse the impurity in the emitter semiconductor electrode 33 into the undoped Si layer by thermal annealing, at the upper-most part of the base 20 to form an n type emitter 35. The remaining part of the base 20 is the intrinsic base layer 25. As a result, a super self-aligned transistor in which the collector-base part is self-aligned through the undercut and in which the emitter-base part is self aligned through the first and the second side-walls is formed without using masks.
Since the depth of the undercut 27a is controlled by the wet etching time, it is difficult to get a good uniformity of the collector-base junction parasitic capacitance which is determined by the depth. When the base layer is selectively grown on the collector layer in the oxide-patterned wafer, the thickness of the base layer, the doping concentration in the base layer, and Ge concentration in the SiGe layer vary according to the density and size of the exposed part of the collector layer. This effect, so called loading effect, can lower process stability and lead to non-uniform device performance within the wafer. In order to lower the loading effect, the growing pressure of the base layer has to be decreased. However, the throughput is lowered in this case. In this prior art, the base electrode is composed with p+ polysilicon. As the parasitic resistance of the p+ polysilicon is bigger than a metal, it obstructs the fmax being increased.
The second related art owned by the IBM in the U.S.A. is about a SiGe HBT with a titanium silicide layer which is used as an ohmic electrode at the emitter, the base and the collector, as shown in FIG. 2, to reduce the parasitic resistances in the emitter, the base and the collector. The method for manufacturing thereof will be briefly given hereinafter.
In FIG. 2a, a buried collector 11 is formed by implanting an n+ type impurity into a p-type silicone substrate 1. A collector layer 10 is formed by depositing a silicon thereon. A trench is formed by etching the substrate 1 and the collector layer 10 for device isolation. An insulation layer is formed with insulation material like a silicon oxide at the inner, side of the trench. A polysilicon is filled into the rest of the inner side of the trench and planarized by chemical-mechanical polishing. As a result, the polysilicon-filled trench isolation 71 is formed. A collector insulation layer (field oxide film) 17 is formed by the recessed local oxidation of silicon (recessed LOCOS). In this method, the part excluding the active region of the collector layer 10 is etched to a pre-determined depth and the part of the collector layer excluding already etched part is thermally oxidized. In other words, the collector insulation layer 17 is thermally oxidized. In other words, the collector insulation layer 17 is formed only in the region of the collector layer 10 excluding the collector sinker 13 which will be formed and the collector 15 on which emitter will be formed. An n+ type collector sinker 13 is formed by implanting n type impurity ions into a part of the collector layer 10 using a photo-resistor and the collector insulation layer 37 as masks. On the resulting structure, a base layer including a p+ SiGe layer and undoped Si layer is grown at the top of the base layer which becomes an emitter at the diffusing of an n type impurity afterwards. At this step, mono-crystalline base layer to be used as a base 25 is deposited on the active collector 15 while poly-crystalline or amorphous base layer to be used as the base semiconductor electrode 23 is deposited on the collector insulation layer 17. The outer part of the base semiconductor electrode is etched off using a photo-resistor mask. An emitter insulation layer 37 like a silicon oxide is deposited. An emitter region is opened by eliminating the part corresponding to the active collector 15 and the base 25 of the emitter insulation layer 37 using a photo resistor mask.
In FIG. 2b, an emitter semiconductor electrode 33 is formed by depositing and patterning an n+ polysilicon. A side-wall silicon oxide film 77 is formed at the outer side of the emitter semiconductor electrode 33 by depositing silicon oxide material and etching it anisotropically. By further anisotropic oxide etching, the emitter insulation layer 37 on the base semiconductor electrode 23 is eliminated.
Using a selective titanium silicide formation only on the silicon surface, a collector ohmic electrode 19, a base ohmic electrode 29, and an emitter ohmic electrode 39 are formed simultaneously. In the present device, the contact resistance and the base parasitic resistance is reduced by the metallic silicide ohmic electrodes 19, 29, 39 on the semiconductor electrodes.
In this case, in order for the emitter metal electrode to be formed on the emitter ohmic electrode, the size of the emitter ohmic electrode should be bigger than that of the emitter metal electrode. The narrower the emitter is, the higher the fT and the fmax are. Therefore the edge of the emitter should be apart from the edge of the base ohmic electrode to a distance L. In other words, the edge of the intrinsic base beneath the emitter is apart from the edge of the base ohmic electrode in the distance of L. Accordingly, occurrence of the parasitic resistance at the extrinsic base region with the length of L is unavoidable. In order to reduce the emitter contact resistance, emitter ohmic electrode should be larger, but the length of L becomes larger. For this reason, there is difficulty in scaling down the device for higher speed and lower power consumption. Extending the emitter ohmic electrode to the outside of the active device region in order for the emitter metal electrode to be contacted thereon could be one way of overcoming this difficulty. In this case the parasitic emitter resistance occurres at the extended part of the emitter ohmic electrode.
The titanium disilicide base ohmic electrode formed by sputtering the titanium and by reacting it with the silicon underneath agglomerates during the silicide formation so, the titanium disilicide layer may penetrate through the thin base layer and contact the collector layer and the cutoff frequency fT may be lowered. For this reason, it is risky to make the base layer thin unconditionally to achieve high fT. In another way, the base ohmic layer should be formed only in the outside of the active device region at the expense of larger L. As a result, the base parasitic resistance increases and the device performance may be degraded.
In the third SiGe base bipolar device of FIG. 3 presented by TEMIC in Germany, a titanium disilicide layer 29 is used as a base ohmic electrode and self-aligned with an emitter semiconductor electrode 33. The manufacturing method is described briefly in the followings.
In FIG. 3a, a buried collector 11 is formed by implanting n-type impurity ions into a p-type silicon substrate 1. A collector layer is deposited thereon. A collector insulation layer 17 is formed by LOCOS process. The layer 17 is not formed in an active collector region 15 and a collector sinker 13. A base layer with p+ type SiGe and an emitter layer with an n-type silicon are sequentially deposited thereon. A part of this layer deposited on the active collector region 15 is a single crystalline while the other part of it deposited on the collector insulation layer (field oxide layer) 17 is either a poly-crystalline or amorphous. An emitter insulation layer 37 with silicon oxide and a silicon nitride layer are sequentially deposited on the resulting structure. Then, a masking film 91 is formed by patterning the silicon nitride layer using a photo-mask which covers the emitter region. The n-type silicon emitter layer outside the emitter region is converted to a p++ type first base semiconductor electrode film 21a By implanting BF2 ion thereon and annealing. The n-type silicon emitter layer inside the emitter region, i.e. intrinsic emitter layer 35, remains unchanged. At the same time, the p+SiGe base layer outside the masking layer becomes a p++ second base semiconductor electrode film 21b while the p+ SiGe base layer beneath the intrinsic emitter layer 35, i.e. intrinsic base layer 25, remains unchanged. The B(boron) diffuses and thereby a pedestal p++ region 27 is formed along with the inner periphery of the active collector region 15, during the thermal annealing which follows the BF2 implantation.
In FIG. 3b, a first and a second base semiconductor electrode, 23a and 23b respectively, are completely formed by patterning the films 21a and 21b using a photo-mask which defines the base electrode region. A first insulating side-wall 73 is formed at the side-walls of the nitride masking film 91 and the base semiconductor electrodes consisting of 23a and 23b, by depositing a silicon oxidation layer and anisotropically etching. A part of emitter insulation layer 37 uncovered by the masking film 91 and the first insulating side-wall are further eliminated by the anisotropic etching. Then, a titanium-silicide base ohmic electrode 29 and collector ohmic electrode 19 are formed selectively only on the upper part of the exposed first semiconductor electrode 23a and the collector sinker 13. A protection layer 79 like a silicon oxide layer is deposited on the resulting substrate. The protection layer 79 is planarized by a chemical mechanical polishing(CMP) until the top surface of the masking film 91 is exposed.
The masking layer 91 is selectively etched off. A second side insulation layer 75 is formed at the inner side of the first side insulation layer 73. A part of the emitter insulation layer 37 is hence exposed and then eliminated by etching so that the n-type silicon emitter layer is opened. An emitter semiconductor electrode 33 is formed by depositing and patterning an n+ type poly-crystalline silicon using a photo-mask. An emitter ohmic electrode 39 is selectively formed with a titanium disilicide only on the emitter semiconductor electrode 33, by sputtering a titanium and thermal annealing. Metal contact windows of the base ohmic electrode 29 and the collector ohmic electrode 19 are opened by patterning the protection layer 79 using a photo-mask. A base terminal 81, an emitter terminal 83, and a collector terminal 85 are formed thereon by depositing and patterning the metal using a photo-mask.
In this method, at the thermal annealing step to activate the boron ions implanted in the base semiconductor electrode 23a and 23b, the boron ions in the p+ SiGe intrinsic base layer 25 diffuses vertically into the adjacent silicon layers; the intrinsic emitter layer 35 and the active collector layer 15. Therefore the base is thickened and cut-off frequency is lowered accordingly. At the same time, the boron implanted in the base semiconductor electrode 23a and 23b laterally diffuses and contacts the n type impurity which diffuses from the n+ type poly-crystalline silicon emitter semiconductor electrode 33. In this case, an n+/p++ emitter/base junction is made and a tunneling current, in other words, a leakage current occurs.
Since a polishing rate of a silicon oxide is almost the same to that of a silicon nitride in planarizing the protection layer 79 at the CMP step, it is difficult to stop the polishing process of the protection layer 79 when the top surface of the masking film 91 is exposed. The first insulating side-wall 73 may be over-polished and eliminate with the protection layer 79. Therefore, the base ohmic electrode 29 maybe in contact with the emitter semiconductor electrode 33. In addition, as the protection layer 79 outside the protruding part is also polished at the same time, it could be eliminated and hence the emitter semiconductor electrode 33 to be formed later could be in contact with the titanium disilicide base ohmic electrode 29. In order to avoid this kind of difficulty such as over-polishing of first insulating side-wall 73 and the protection layer 79 outside the protruding part, it is desirable to make the masking film 91 thicker. However, a thick silicon nitride film puts on lots of stress to the substrate when it is deposited. Therefore, there is little process control margin.
The fourth related art, as shown in FIG. 4, is a SiGe base bipolar transistor proposed by ETRI in Korea. The method for manufacturing thereof will be described briefly in the followings.
In FIG. 4a, an n+ buried collector 11 is formed by ion-implanting n type impurities into a pxe2x88x92 type silicon substrate 1. A collector layer is formed with n-type silicon thereon. A collector insulation layer 17 is formed by applying LOCOS process to the region in which a collector active region 15 and a collector sinker 13 are not formed. An n+ collector sinker 13 is formed by implanting an n type impurity into a collector sinker region. A base layer 20 composed of three layers, an undoped SiGe, a p+ SiGe, and undoped Si layer from the bottom to the top, is deposited on the entire substrate 1 where the active collector region 15 and the collector sinker 13 are formed. At this process step, the crystallinity of the base layer 20 deposited on the active collector 15 and the collector sinker 13 is a single crystalline, while that on the collector insulation layer (field oxidation film) 17 is a poly-crystalline or an amorphous. A masking film 91 covering the collector sinker 13 and an active base region 25 within the active collector region 15 is formed by depositing and patterning a silicon oxide film on the base layer. The BF2 ions are implanted to the exposed base layer using the masking film 91 as a mask. A heat treatment is followed in order to activate the implanted ions and re-crystallize the damaged silicon layer by the implanting. At the same time, the implanted impurity ions diffuse so that a p++ base semiconductor electrode film 21 and a p++ region 27 at the edge of the collector 15 region are formed. Then the masking film 91 is eliminated. An amorphous base ohmic electrode film 29 is deposited on the base layer by sputtering a composite metal like TiSi2.6. A silicon oxide film 93 is further deposited on the base ohmic electrode film 29. A part of the silicon oxide film 93 on the active base 25 is etched using a photo-mask which opens the active base region 15. Then, the opened part of the base ohmic electrode film 29 is wet etched accordingly by a HF-base chemical solution. An emitter insulation layer 37 like a silicon oxide layer is deposited on the resulting substrate. The emitter insulation layer 37 and the silicon oxide film 93 are etched using a photomask covering the base electrode region. Then, the base ohmic electrode film 29 is wet-etched and the base layer 20 on collector insulation film 17 is dry-etched using the remaining emitter insulation layer 37 and the remaining silicon oxide film 93 as masks. A side-wall insulation layer 77 is formed at each of the etched side-wall of the emitter insulation layer 37, the silicon oxide film 93, the base ohmic electrode film 29, and the base semiconductor electrode 23. The emitter region is opened using a photomask by etching the emitter insulation layer 37 and consequently the active base layer 25 is exposed. An emitter semiconductor electrode 33 and a collector semiconductor electrode 13a are simultaneously formed on the active base layer 25 and the collector sinker 13 by depositing a polysilicon layer on the entire resulting substrate, implanting an n type impurity into the polysilicon layer, and patterning the polysilicon layer using a photomask. A passivation layer 79 is formed on the entire resulting substrate by depositing a silicon oxide. Then, by thermal annealing the substrate, the n type impurity in the emitter semiconductor electrode 33 is diffused into the adjacent layer so that the undoped Si layer, which is the upper layer of the active base layer, is converted to an emitter 35. By patterning the passivation layer 70 using a photomask, metal contact windows for the emitter, the base, and the collector are formed on the emitter semiconductor electrode 33, the base ohmic electrode 29, and collector semiconductor electrode 13a, respectively. Then, a metal film including TiW and A-1%Si is deposited and patterned using a phtomask defining metal interconnection. Consequently, an emitter terminal 83, a base terminal 81, and a collector terminal 85 are formed.
This related art also has drawbacks like other previously cited arts. First, when wet etching the amorphous TiSi2.6 base ohmic electrode film 29 using a photo-mask which opens the active base region 15, the wet etch rate of the amorphous TiSi2.6 base ohmic electrode film 29 is different from that of the silicon oxide film 93. Consequently, the side-wall of the etched active base region becomes so uneven that a void generation at the uneven side-wall is resulted when depositing the emitter insulation layer 37. When the polysilicon layer is deposited to be the emitter semiconductor electrode 33 later on, the void is filled with the polysilicon and therefore the emitter semiconductor electrode 33 is in contact with the base ohmic electrode. In addition, since the etching process inevitably introduces bubbles at the reacting surface of the amorphous TiSi2.6 base ohmic electrode film 29, the bubbles prohibit the wet etchant from etching the amorphous TiSi2.6 film. As a result, parts of the amorphous TiSi2.6 film covered with the bubbles are etched insufficiently that the amorphous TiSi2.6 residues may cause electrical contact between the emitter semiconductor electrode 33 and the base ohmic electrode film 29.
Second, when wet-etching the base ohmic electrode film 29 outside of the base electrode region using the emitter insulation layer 37 and the silicon oxide film 93 as masks, the difference of the etch rate difference between the silicon oxide and the amorphous TiSi2.6 film results in the uneven side-wall again. When cleaning the resulting wafer using a diluted HF, the emitter insulation layer 37, the silicon oxide film 93, and the amorphous TiSi2.6 film 29 are further etched so that the base electrode boundary become more unclear. That is, the base electrode formation process is not stable and therefore the resulting base electrode becomes degraded electrically.
As forementioned, there are drawbacks and problems in the cited arts of the SiGe base bipolar transistor. In order to achieve higher cutoff frequency in a bipolar transistor, it is appropriate to have the base layer as thin as possible. Since a titanium disilicide is likely to agglomerate, if the ultra thin base bipolar transistors is formed with the titanium base ohmic electrode, the titanium disilicide is likely to permeate the thin base layer and thus contact the collector layer. In order to avoid this problem, the titanium base ohmic electrode is formed on the base layer outside the active collector region in the prior art of FIG. 2 at the expense of longer L as illustrated in FIG. 2a. Another way is suggested in FIG. 4 to overcome this drawback. When wet etching the base electrode formed with amorphous TiSi2.6, uniformity in the processing step cannot be guaranteed and an electrical contact is likely to occur between the base and the emitter due to the amorphous TiSi2.6 residues at the etched surface of the active base region.
The object of the present invention is to provide an epitaxially grown Si- or SiGe-base bipolar transistor having a metallic silicide base ohmic layer to reduce the base electrode resistance and a manufacturing the method thereof without aforementioned drawbacks. According to the present invention, a homojunction or heterojunction bipolar transistor comprises of a semiconductor substrate having a buried collector; an active collector and a collector sinker thereon which are divided by a collector insulation film; a base layer divided into an active base layer on the active collector and a first base semiconductor electrode film on the collector insulation film; a second base semiconductor electrode film selectively grown only on the first base semiconductor electrode film by using a masking film; a metallic silicide base ohmic electrode film selectively formed on the second base semiconductor electrode film; an emitter insulation film isolating an emitter; an emitter formed on the active base layer; an emitter semiconductor electrode formed on the emitter; a passivation film entirely covering the resulting structure; an emitter electrode, a base electrode, and a collector electrode formed on the emitter semiconductor electrode, the base ohmic electrode, and the collector sinker, respectively.